Classification Model for Car Insurance Claims¶

In [1]:
!pip install scikit-learn

Requirement already satisfied: scikit-learn in c:\users\westo\anaconda3\lib\site-packages (1.3.0)
Requirement already satisfied: numpy>=1.17.3 in c:\users\westo\anaconda3\lib\site-packages (from scikit-learn) (1.24.3)
Requirement already satisfied: scipy>=1.5.0 in c:\users\westo\anaconda3\lib\site-packages (from scikit-learn) (1.10.1)
Requirement already satisfied: joblib>=1.1.1 in c:\users\westo\anaconda3\lib\site-packages (from scikit-learn) (1.2.0)
Requirement already satisfied: threadpoolctl>=2.0.0 in c:\users\westo\anaconda3\lib\site-packages (from scikit-learn) (2.2.0)
In [2]:
import pandas as pd
import numpy as np 
import seaborn as sns
from matplotlib import pyplot as plt
import math

Section 1: Data Pre-Processing¶

In [3]:
#Convert csv to dataframe

df = pd.read_csv("car_insurance_claim.csv")
In [4]:
#A look at the dataframe

df.head()
Out[4]:
ID KIDSDRIV BIRTH AGE HOMEKIDS YOJ INCOME PARENT1 HOME_VAL MSTATUS ... CAR_TYPE RED_CAR OLDCLAIM CLM_FREQ REVOKED MVR_PTS CLM_AMT CAR_AGE CLAIM_FLAG URBANICITY
0 63581743 0 16MAR39 60.0 0 11.0 $67,349 No $0 z_No ... Minivan yes $4,461 2 No 3 $0 18.0 0 Highly Urban/ Urban
1 132761049 0 21JAN56 43.0 0 11.0 $91,449 No $257,252 z_No ... Minivan yes $0 0 No 0 $0 1.0 0 Highly Urban/ Urban
2 921317019 0 18NOV51 48.0 0 11.0 $52,881 No $0 z_No ... Van yes $0 0 No 2 $0 10.0 0 Highly Urban/ Urban
3 727598473 0 05MAR64 35.0 1 10.0 $16,039 No $124,191 Yes ... z_SUV no $38,690 2 No 3 $0 10.0 0 Highly Urban/ Urban
4 450221861 0 05JUN48 51.0 0 14.0 NaN No $306,251 Yes ... Minivan yes $0 0 No 0 $0 6.0 0 Highly Urban/ Urban

5 rows × 27 columns

In [5]:
#Remove ID, Birthdate, and Claim Amount columns as they are either irrelevant or too indicative

df = df.drop(columns=['ID', 'CLM_AMT', 'BIRTH'], axis=1)

df.head()
Out[5]:
KIDSDRIV AGE HOMEKIDS YOJ INCOME PARENT1 HOME_VAL MSTATUS GENDER EDUCATION ... TIF CAR_TYPE RED_CAR OLDCLAIM CLM_FREQ REVOKED MVR_PTS CAR_AGE CLAIM_FLAG URBANICITY
0 0 60.0 0 11.0 $67,349 No $0 z_No M PhD ... 11 Minivan yes $4,461 2 No 3 18.0 0 Highly Urban/ Urban
1 0 43.0 0 11.0 $91,449 No $257,252 z_No M z_High School ... 1 Minivan yes $0 0 No 0 1.0 0 Highly Urban/ Urban
2 0 48.0 0 11.0 $52,881 No $0 z_No M Bachelors ... 1 Van yes $0 0 No 2 10.0 0 Highly Urban/ Urban
3 0 35.0 1 10.0 $16,039 No $124,191 Yes z_F z_High School ... 4 z_SUV no $38,690 2 No 3 10.0 0 Highly Urban/ Urban
4 0 51.0 0 14.0 NaN No $306,251 Yes M <High School ... 7 Minivan yes $0 0 No 0 6.0 0 Highly Urban/ Urban

5 rows × 24 columns

In [6]:
sns.countplot(x="CLAIM_FLAG", data=df)
Out[6]:
<Axes: xlabel='CLAIM_FLAG', ylabel='count'>
In [7]:
df.isna().sum()
Out[7]:
KIDSDRIV        0
AGE             7
HOMEKIDS        0
YOJ           548
INCOME        570
PARENT1         0
HOME_VAL      575
MSTATUS         0
GENDER          0
EDUCATION       0
OCCUPATION    665
TRAVTIME        0
CAR_USE         0
BLUEBOOK        0
TIF             0
CAR_TYPE        0
RED_CAR         0
OLDCLAIM        0
CLM_FREQ        0
REVOKED         0
MVR_PTS         0
CAR_AGE       639
CLAIM_FLAG      0
URBANICITY      0
dtype: int64
In [8]:
df.info()

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 10302 entries, 0 to 10301
Data columns (total 24 columns):
 #   Column      Non-Null Count  Dtype  
---  ------      --------------  -----  
 0   KIDSDRIV    10302 non-null  int64  
 1   AGE         10295 non-null  float64
 2   HOMEKIDS    10302 non-null  int64  
 3   YOJ         9754 non-null   float64
 4   INCOME      9732 non-null   object 
 5   PARENT1     10302 non-null  object 
 6   HOME_VAL    9727 non-null   object 
 7   MSTATUS     10302 non-null  object 
 8   GENDER      10302 non-null  object 
 9   EDUCATION   10302 non-null  object 
 10  OCCUPATION  9637 non-null   object 
 11  TRAVTIME    10302 non-null  int64  
 12  CAR_USE     10302 non-null  object 
 13  BLUEBOOK    10302 non-null  object 
 14  TIF         10302 non-null  int64  
 15  CAR_TYPE    10302 non-null  object 
 16  RED_CAR     10302 non-null  object 
 17  OLDCLAIM    10302 non-null  object 
 18  CLM_FREQ    10302 non-null  int64  
 19  REVOKED     10302 non-null  object 
 20  MVR_PTS     10302 non-null  int64  
 21  CAR_AGE     9663 non-null   float64
 22  CLAIM_FLAG  10302 non-null  int64  
 23  URBANICITY  10302 non-null  object 
dtypes: float64(3), int64(7), object(14)
memory usage: 1.9+ MB
In [9]:
def fix_money(word):
    if not pd.isnull(word):
        new_money = word.replace(",", "")
        return float(new_money[1:])
    else:
        return word
In [10]:
money_list = ["INCOME", "HOME_VAL", "BLUEBOOK", "OLDCLAIM"]

for i in money_list:
    df[i] = df[i].apply(fix_money)
    
df
Out[10]:
KIDSDRIV AGE HOMEKIDS YOJ INCOME PARENT1 HOME_VAL MSTATUS GENDER EDUCATION ... TIF CAR_TYPE RED_CAR OLDCLAIM CLM_FREQ REVOKED MVR_PTS CAR_AGE CLAIM_FLAG URBANICITY
0 0 60.0 0 11.0 67349.0 No 0.0 z_No M PhD ... 11 Minivan yes 4461.0 2 No 3 18.0 0 Highly Urban/ Urban
1 0 43.0 0 11.0 91449.0 No 257252.0 z_No M z_High School ... 1 Minivan yes 0.0 0 No 0 1.0 0 Highly Urban/ Urban
2 0 48.0 0 11.0 52881.0 No 0.0 z_No M Bachelors ... 1 Van yes 0.0 0 No 2 10.0 0 Highly Urban/ Urban
3 0 35.0 1 10.0 16039.0 No 124191.0 Yes z_F z_High School ... 4 z_SUV no 38690.0 2 No 3 10.0 0 Highly Urban/ Urban
4 0 51.0 0 14.0 NaN No 306251.0 Yes M <High School ... 7 Minivan yes 0.0 0 No 0 6.0 0 Highly Urban/ Urban
... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ...
10297 1 45.0 2 9.0 164669.0 No 386273.0 Yes M PhD ... 15 Minivan no 0.0 0 No 2 17.0 0 Highly Urban/ Urban
10298 0 46.0 0 9.0 107204.0 No 332591.0 Yes M Masters ... 6 Panel Truck no 0.0 0 No 0 1.0 0 Highly Urban/ Urban
10299 0 48.0 0 15.0 39837.0 No 170611.0 Yes z_F <High School ... 7 z_SUV no 0.0 0 No 0 1.0 0 Highly Urban/ Urban
10300 0 50.0 0 7.0 43445.0 No 149248.0 Yes z_F Bachelors ... 6 Minivan no 0.0 0 No 0 11.0 0 Highly Urban/ Urban
10301 0 52.0 0 11.0 53235.0 No 197017.0 Yes z_F z_High School ... 6 Minivan no 0.0 0 No 0 9.0 0 z_Highly Rural/ Rural

10302 rows × 24 columns

In [11]:
null_list = ["YOJ", "INCOME", "HOME_VAL", "CAR_AGE"]

for i in null_list:
    df[i] = df[i].fillna(
        df[i].mean()
    )
In [12]:
for i in df.columns[:-1]:
    plt.figure(i)
    sns.countplot(x = i, data=df, hue = 'CLAIM_FLAG')

C:\Users\westo\AppData\Local\Temp\ipykernel_48564\816024472.py:2: RuntimeWarning: More than 20 figures have been opened. Figures created through the pyplot interface (`matplotlib.pyplot.figure`) are retained until explicitly closed and may consume too much memory. (To control this warning, see the rcParam `figure.max_open_warning`). Consider using `matplotlib.pyplot.close()`.
  plt.figure(i)
In [13]:
len(df)
Out[13]:
10302
In [14]:
df = df.dropna(axis=0)
In [15]:
len(df)
Out[15]:
9630
In [16]:
df.info()

<class 'pandas.core.frame.DataFrame'>
Int64Index: 9630 entries, 0 to 10301
Data columns (total 24 columns):
 #   Column      Non-Null Count  Dtype  
---  ------      --------------  -----  
 0   KIDSDRIV    9630 non-null   int64  
 1   AGE         9630 non-null   float64
 2   HOMEKIDS    9630 non-null   int64  
 3   YOJ         9630 non-null   float64
 4   INCOME      9630 non-null   float64
 5   PARENT1     9630 non-null   object 
 6   HOME_VAL    9630 non-null   float64
 7   MSTATUS     9630 non-null   object 
 8   GENDER      9630 non-null   object 
 9   EDUCATION   9630 non-null   object 
 10  OCCUPATION  9630 non-null   object 
 11  TRAVTIME    9630 non-null   int64  
 12  CAR_USE     9630 non-null   object 
 13  BLUEBOOK    9630 non-null   float64
 14  TIF         9630 non-null   int64  
 15  CAR_TYPE    9630 non-null   object 
 16  RED_CAR     9630 non-null   object 
 17  OLDCLAIM    9630 non-null   float64
 18  CLM_FREQ    9630 non-null   int64  
 19  REVOKED     9630 non-null   object 
 20  MVR_PTS     9630 non-null   int64  
 21  CAR_AGE     9630 non-null   float64
 22  CLAIM_FLAG  9630 non-null   int64  
 23  URBANICITY  9630 non-null   object 
dtypes: float64(7), int64(7), object(10)
memory usage: 1.8+ MB
In [17]:
sns.pairplot(df, hue="CLAIM_FLAG")
Out[17]:
<seaborn.axisgrid.PairGrid at 0x1bb50feb650>
In [18]:
from sklearn.preprocessing import LabelEncoder

#Encode non-numeric data

obj_list = ["PARENT1", "MSTATUS", "GENDER", "EDUCATION", "OCCUPATION", "CAR_USE", "CAR_TYPE", "RED_CAR", "REVOKED", "URBANICITY"]

for i in obj_list:
    le = LabelEncoder()
    le.fit(df[i])
    df[i] = le.transform(df[i])

C:\Users\westo\AppData\Local\Temp\ipykernel_48564\662962883.py:10: SettingWithCopyWarning: 
A value is trying to be set on a copy of a slice from a DataFrame.
Try using .loc[row_indexer,col_indexer] = value instead

See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy
  df[i] = le.transform(df[i])
C:\Users\westo\AppData\Local\Temp\ipykernel_48564\662962883.py:10: SettingWithCopyWarning: 
A value is trying to be set on a copy of a slice from a DataFrame.
Try using .loc[row_indexer,col_indexer] = value instead

See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy
  df[i] = le.transform(df[i])
C:\Users\westo\AppData\Local\Temp\ipykernel_48564\662962883.py:10: SettingWithCopyWarning: 
A value is trying to be set on a copy of a slice from a DataFrame.
Try using .loc[row_indexer,col_indexer] = value instead

See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy
  df[i] = le.transform(df[i])
C:\Users\westo\AppData\Local\Temp\ipykernel_48564\662962883.py:10: SettingWithCopyWarning: 
A value is trying to be set on a copy of a slice from a DataFrame.
Try using .loc[row_indexer,col_indexer] = value instead

See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy
  df[i] = le.transform(df[i])
C:\Users\westo\AppData\Local\Temp\ipykernel_48564\662962883.py:10: SettingWithCopyWarning: 
A value is trying to be set on a copy of a slice from a DataFrame.
Try using .loc[row_indexer,col_indexer] = value instead

See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy
  df[i] = le.transform(df[i])
C:\Users\westo\AppData\Local\Temp\ipykernel_48564\662962883.py:10: SettingWithCopyWarning: 
A value is trying to be set on a copy of a slice from a DataFrame.
Try using .loc[row_indexer,col_indexer] = value instead

See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy
  df[i] = le.transform(df[i])
C:\Users\westo\AppData\Local\Temp\ipykernel_48564\662962883.py:10: SettingWithCopyWarning: 
A value is trying to be set on a copy of a slice from a DataFrame.
Try using .loc[row_indexer,col_indexer] = value instead

See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy
  df[i] = le.transform(df[i])
C:\Users\westo\AppData\Local\Temp\ipykernel_48564\662962883.py:10: SettingWithCopyWarning: 
A value is trying to be set on a copy of a slice from a DataFrame.
Try using .loc[row_indexer,col_indexer] = value instead

See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy
  df[i] = le.transform(df[i])
C:\Users\westo\AppData\Local\Temp\ipykernel_48564\662962883.py:10: SettingWithCopyWarning: 
A value is trying to be set on a copy of a slice from a DataFrame.
Try using .loc[row_indexer,col_indexer] = value instead

See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy
  df[i] = le.transform(df[i])
C:\Users\westo\AppData\Local\Temp\ipykernel_48564\662962883.py:10: SettingWithCopyWarning: 
A value is trying to be set on a copy of a slice from a DataFrame.
Try using .loc[row_indexer,col_indexer] = value instead

See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy
  df[i] = le.transform(df[i])

Section 2: Model Creation¶

Testing Different Models¶

In [19]:
from sklearn.linear_model import LogisticRegression
from sklearn.model_selection import train_test_split
from sklearn.inspection import permutation_importance

X = df.drop(columns=['CLAIM_FLAG'])
t = df['CLAIM_FLAG']

X_train, X_test, t_train, t_test = train_test_split(X, t, test_size=0.2, random_state=0)
In [20]:
clf = LogisticRegression()

clf.fit(X_train, t_train)

train_score = clf.score(X_train, t_train)

test_score = clf.score(X_test, t_test)

model_fi = permutation_importance(clf, X_train, t_train)
imp_dict = {}

for i in range(23):
    imp_dict[X_train.columns[i]] = model_fi['importances_mean'][i]
    
imp_sorted = dict(sorted(imp_dict.items(), key=lambda item: item[1]))    

print(imp_sorted)

plt.figure(figsize=(25, 5))  # width:20, height:3
plt.bar(range(len(imp_sorted)), imp_sorted.values(), align='edge', width=0.4)
plt.xticks(range(len(imp_sorted)), list(imp_sorted.keys()))

print("Train Accuracy: {}, Test Accuracy: {}".format(train_score, test_score))

{'BLUEBOOK': -0.0006230529595015577, 'AGE': -0.0004153686396677125, 'CAR_AGE': -0.00031152647975076777, 'MVR_PTS': -0.00020768431983384517, 'TRAVTIME': -0.00015576323987538389, 'YOJ': -0.00012980269989615323, 'CLM_FREQ': -5.192107995846129e-05, 'CAR_TYPE': -2.5960539979230646e-05, 'INCOME': -2.2204460492503132e-17, 'KIDSDRIV': 0.0, 'HOMEKIDS': 0.0, 'PARENT1': 0.0, 'MSTATUS': 0.0, 'GENDER': 0.0, 'EDUCATION': 0.0, 'CAR_USE': 0.0, 'RED_CAR': 0.0, 'REVOKED': 0.0, 'URBANICITY': 0.0, 'OCCUPATION': 5.192107995846129e-05, 'TIF': 0.00033748701972999837, 'HOME_VAL': 0.0012720664589823461, 'OLDCLAIM': 0.0062045690550363245}
Train Accuracy: 0.7316978193146417, Test Accuracy: 0.7320872274143302
In [21]:
X = df[['AGE', 'YOJ', 'HOME_VAL', 'TRAVTIME', 'TIF', 'OLDCLAIM', 'CLM_FREQ', 'MVR_PTS', 'CAR_AGE']]
t = df['CLAIM_FLAG']

X_train, X_test, t_train, t_test = train_test_split(X, t, test_size=0.2, random_state=0)
In [22]:
clf = LogisticRegression()

clf.fit(X_train, t_train)

train_score = clf.score(X_train, t_train)

test_score = clf.score(X_test, t_test)

print("Train Accuracy: {}, Test Accuracy: {}".format(train_score, test_score))

Train Accuracy: 0.7472741433021807, Test Accuracy: 0.7440290758047767

C:\Users\westo\anaconda3\Lib\site-packages\sklearn\linear_model\_logistic.py:460: ConvergenceWarning: lbfgs failed to converge (status=1):
STOP: TOTAL NO. of ITERATIONS REACHED LIMIT.

Increase the number of iterations (max_iter) or scale the data as shown in:
    https://scikit-learn.org/stable/modules/preprocessing.html
Please also refer to the documentation for alternative solver options:
    https://scikit-learn.org/stable/modules/linear_model.html#logistic-regression
  n_iter_i = _check_optimize_result(
In [23]:
from sklearn.naive_bayes import GaussianNB
clf = GaussianNB()

clf.fit(X_train, t_train)

train_score = clf.score(X_train, t_train)

test_score = clf.score(X_test, t_test)

print("Train Accuracy: {}, Test Accuracy: {}".format(train_score, test_score))

Train Accuracy: 0.7306593977154725, Test Accuracy: 0.7289719626168224
In [24]:
from sklearn.linear_model import Ridge

clf = Ridge()

clf.fit(X_train, t_train)

train_score = clf.score(X_train, t_train)

test_score = clf.score(X_test, t_test)

print("Train Accuracy: {}, Test Accuracy: {}".format(train_score, test_score))

Train Accuracy: 0.11693644948376292, Test Accuracy: 0.10273547986118592
In [25]:
from sklearn.neighbors import KNeighborsClassifier

clf = KNeighborsClassifier()

clf.fit(X_train, t_train)

train_score = clf.score(X_train, t_train)

test_score = clf.score(X_test, t_test)

print("Train Accuracy: {}, Test Accuracy: {}".format(train_score, test_score))

Train Accuracy: 0.7940031152647975, Test Accuracy: 0.7087227414330218
In [26]:
from sklearn.tree import DecisionTreeClassifier

clf = DecisionTreeClassifier()

clf.fit(X_train, t_train)

train_score = clf.score(X_train, t_train)

test_score = clf.score(X_test, t_test)

print("Train Accuracy: {}, Test Accuracy: {}".format(train_score, test_score))

Train Accuracy: 0.9996105919003115, Test Accuracy: 0.6490134994807892
In [27]:
from sklearn.metrics import confusion_matrix, classification_report, ConfusionMatrixDisplay

clf = LogisticRegression(max_iter=10000)

clf.fit(X_train, t_train)

preds = clf.predict(X_test)

cm_report = classification_report(t_test, preds)

print(cm_report)

cm = confusion_matrix(t_test, preds)
disp = ConfusionMatrixDisplay(confusion_matrix=cm)
disp.plot()
plt.show()

              precision    recall  f1-score   support

           0       0.76      0.95      0.85      1416
           1       0.58      0.18      0.28       510

    accuracy                           0.75      1926
   macro avg       0.67      0.57      0.56      1926
weighted avg       0.71      0.75      0.70      1926